In the production of semiconductors, such as memory, processors, and controllers, among others, a mask is used. The mask is placed over a semiconductor wafer to expose or shield different portions of the wafer from light, or some other element. The exposed wafer is then processed with etching, deposition and other processes to produce the features of the various semiconductors in the wafer that make up the finished product.
The masks are designed using computer design programs that derive an aerial view or image of the wafer based on the electronic circuitry that is to be built on the wafer. The mask is designed to produce this aerial image on the wafer based on using a particular set of photolithography equipment. In other words, the mask must be designed so that when a particular wavelength of light at a particular distance is directed to a wafer through a particular set of optics and the mask, the desired pattern will be illuminated with the desired intensity on the wafer.
The pattern on the mask may be very complex and finely detailed. In some systems, a mask is designed with a matrix of pixels in columns and rows that illuminate a wafer that has an area of about one square centimeter. The mask may be four or more times that size and reduction optics are used to reduce the mask image down to the size of the wafer. For a 193 nm light source, each pixel may be about 100 nm across so that the mask may have billions of pixels. Each pixel is either a transparent spot on the mask (1), an opaque spot on the mask (0), or a transparent spot that reverses the phase of the light passing through (−1). The use of three different values (+1, 0, −1) allows for greater control over the diffractive effects through the mask.
In order to enhance the accuracy and the resolution of the pattern that results on the wafer. A variety of different optimization techniques are typically applied to the mask. One such technique is to add sub-resolution assist features (SRAF) or scattering bars to a mask. These are usually small features in the form of parallel lines or spaces that are smaller than the resolution limit of the imaging system. In other words, the features are too small to be printed on the wafer through the lens but they influence the lithographic behavior of the larger features that they are near. For example, SRAFs in the form of parallel lines running along either side of a solid line improves the focus of the solid line.
SRAFs can be used to ensure that features will be printed correctly on the wafer even as the parameters of the printing process (focus, intensity, chemistry, wafer composition, etc.) vary through their anticipated range. (The combination of these variations of the parameters of the printing process are sometimes referred to as the process window.) SRAFs have been combined with optical proximity correction (OPC), off-axis illumination (OAI), attenuated phase shifted mask (APSM) enhanced lithography, embedded phase shifted mask (EPSM) lithography, and other techniques for even more accurate photolithography.
However, for non-collinear structures, i.e. structures that are not aligned along the same line, SRAFs cannot be used between the structures. This makes it more difficult to pattern non-collinear structures that are very close together. When structures on the wafer are to be printed very close together, the mask error enhancement factor (MEEF or mask error factor MEF) tends to increase which, in turn, leads to high variations in the critical dimension (CD) for the process across the printed area. The MEEF represents how much the size of a feature printed on a mask changes in response to a change in the mask. A MEEF of 1.0 indicates that a change in the mask causes a proportional change in the final printed wafer. In other words, moving a line 4 micrometers in the mask will move the same line 1 micrometer in the printed wafer, if the lithography optics reduces the mask image on the wafer by a factor of four. When features become small enough to be near the resolution limit of the photolithography system, the MEEF increases dramatically. This means that a small change on the mask produces a very large change on the printed wafer. This makes it difficult to precisely control feature sizes. The mask design is also made more complex because different features on the same mask will have different MEEFs.
The MEEF can be reduced by using a second mask and printing some features using one mask and other features using another mask. However, this doubles the time and expense of performing the exposure. The MEEF can also be reduced using hammerhead extensions and serifs. However, when the structures are placed closer together high MEEF and CD variability can occur.